Epoxidation Reaction of Unsaturated Hydrocarbons with H2O2 over

DOI: 10.1016/j.mcat.2018.04.011. Hanlu Wang, YiQiang Deng, ... Physical Chemistry Chemical Physics 2016, 18 (1) , 190-196. DOI: 10.1039/C5CP05268H.Mis...
3 downloads 0 Views 652KB Size
Download PDF
J. Phys. Chem. C 2007, 111, 3433-3441

3433

Epoxidation Reaction of Unsaturated Hydrocarbons with H2O2 over Defect TS-1 Investigated by ONIOM Method: Formation of Active Sites and Reaction Mechanisms Weerayuth Panyaburapa, Tanin Nanok, and Jumras Limtrakul* Laboratory for Computational and Applied Chemistry, Department of Chemistry, Faculty of Science, Kasetsart UniVersity, Bangkok 10900, Thailand, and Center of Nanotechnology, Kasetsart UniVersity Research and DeVelopment Institute, Kasetsart UniVersity, Bangkok 10900, Thailand ReceiVed: August 27, 2006; In Final Form: December 11, 2006

The mechanism of alkene oxidation with hydrogen peroxide over the titanium silicalite-1 (TS-1) defect is investigated using a 65T nanocluster, TiSi64O97H74, and calculated at the 9T/65T two-layered ONIOM(B3LYP/ 6-31G(d,p):UFF) level. The intermediate titanium hydroperoxo in the bidentate form, Ti(η2-OOH), occurring through the single-step double proton-transfer mechanism aided by a neighboring silanol group, is proffered as the active species in the oxidation process. It is noted that this species is influenced by the number of water molecules surrounding the active region. The formation of titanium peroxo species, Ti(η2-OO-), consistent with the role of water in hydroperoxo-peroxo interconversion in the TS-1/H2O/H2O2 system, results from the step in which an additional water molecule is introduced into the hydrated Ti(η2-OOH) complex. The step in which oxygen is abstracted during the epoxide formation is determined to be the reaction rate determining step, and is reactive to a number of methyl groups substituted to the active CdC bond of alkene molecules. The evident activation energies for ethylene, propylene, and trans-2-butylene are estimated to be 15.5, 13.6, and 12.2 kcal/mol, respectively. These results agree with the reactivity series of the gas-phase calculations and compare favorably with the known apparent activation energy of 1-hexene of 15.5 ( 1.5 kcal/mol obtained from experiment.

1. Introduction Zeolite is a heterogeneous catalyst which is widely used for environmental and industrial applications.1 The high activity of inserted transition metal atoms, for example Al and Ti, in the MFI framework-structure type are exploited in various chemical processes. Titanium silicalite-1, TS-1, which has a very high catalytic activity and selectivity in oxidation reactions, is a premium catalyst extensively used in the chemical industry.2 In addition, Ti-containing siliceous nanoporous and mesoporous materials have been found to perform well in mild oxidation reactions in the presence of hydrogen peroxide for various organic compounds such as the epoxidation of unsaturated hydrocarbons,3 the conversion of ammonia to hydroxylamine,4 the oxidation of alkanes,5 alcohols,6 the hydroxylation of aromatic compounds,7-8 and ketone ammoximation.9 Although TS-1 and silicalite-1 have the same MFI-type structure, only TS-1 reveals the catalytic activity in oxidation processes. Thus, the presence of Ti atoms in the TS-1 framework is critical in understanding the structure and catalytic activity. The supposition that the Ti atom in the crystal lattice is the most probable active center in facilitating the oxidation process is based on evidence from theoretical10-13 and experimental studies.14-21 However, the difficulty in determining how such a minute amount of titanium atoms are distributed among the 12 crystallographically distinct T sites has persisted for the past two decades. Numerous recent research efforts have been made to observe their preferential sites in the TS-1 framework, and it has been found that the distribution of Ti atoms in the MFI framework is nonrandom but their favored sites are sensitive to the synthesis conditions.21-24 * Corresponding author: email: [email protected].

It is well-known that titanium atoms in the framework of TS-1 are tetravalent or four-coordinate in anhydrous conditions. This structure has been widely used as an active site model of TS117,24-27 for mild epoxidation processes and a range of proposed active species, which include titanium hydroperoxo complexes, in mono- and bidentate manners, and titanyl form (Figure 1), were considered to be involved in oxidation reactions. The diffuse refractance UV-vis (UV-vis DRS) spectra did not, however, contain information regarding the titanyl species.28 Computational studies of small clusters without solvent molecules have shown that the bidentate titanium hydroperoxo complex, Ti(η2-OOH), was the most energetically favorable active species, and it has been used as an active site in computational studies of the epoxidation reaction of ethylene on the titanium silicalite-1.25 This notwithstanding, when two water molecules were included in the model, the monodentate form, Ti(η1-OOH), was preferred.16 Recently Munakata and his co-workers29 proposed an additional active species known as the peroxo-titanosilicalite complex, Ti-O-O-Si (Figure 1), but this was found to be highly unstable when imposed by the zeolite lattice. When TS-1 is exposed to water, experiment shows, however, that the coordination expansion takes place with the hydrolysis of the Ti-O-Si bridges forming Ti-OH and Si-OH groups.30-33 Therefore, the tripodal structure (one Ti-OH group and three Ti-O-Si bridges anchoring the Ti atom to the silica framework) is likely to be the chemically active form in hydrous titanosilicates. Nevertheless, the considerable repulsion between Si-OH and Ti-OH resulting from hydrolysis has been demonstrated theoretically.34 This raises the issue of whether the Si vacancy, commonly detected in silicalite, at the Ti-neighboring site should be considered in the calculations. Wells et al. found

10.1021/jp065544n CCC: $37.00 © 2007 American Chemical Society Published on Web 02/06/2007

3434 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Panyaburapa et al. 2. Models and Methods

Figure 1. Illustration of proposed oxidative active species in the catalytic epoxidation reaction of unsaturated hydrocarbons with H2O2 over TS-1.

that the introduction of Ti located near Si vacancy sites facilitated the active sites formation and have proposed that Ti(η1-OOH) was the active species for propylene epoxidation.35 In our current research, we have undertaken intensive investigations of the reaction mechanisms of the oxidative active site formation and the epoxidation of unsaturated hydrocarbons. The Ti active center located adjacent to the Si vacancy site was modeled as the TS-1 active site as this model is sufficient to study two different competing reaction channels for the oxidative active site formation. The stability of oxidative active species in different configurations as well as in the presence of water molecules is also discussed. Our research concluded with the examination of the effect of molecular chain length on the partial oxidation of small olefins, including ethylene, propylene and trans-2-butylene, by TS-1 and H2O2. To the best of our knowledge, only small cluster models have been used for studying the mechanism of the active site formation, stability, and catalytic activity in epoxidation. Such small models are not able to locate some reaction pathways which may have a lower barrier. It is essential that the effect of the lattice framework be included to obtain a better perception of the reaction mechanism in the real zeolite pores. Several theoretical models, together with the periodic structure calculations, have been proposed to study the interactions in extended systems of crystals or surfaces.36-40 For nanostructured materials, for example zeolites, usually possess hundreds of atoms per unit cell making the use of accurate periodic structure calculations computationally too expensive and even impractical when very large zeolites are concerned. Alternatively, hybrid methods, such as embedded cluster or combined quantum mechanics/molecular mechanics (QM/MM) methods, as well as the more general Our-own-Nlayered Integrated molecular Orbital + molecular Mechanics (ONIOM) method,38-43 have brought a larger system within reach of obtaining accurate results. The ONIOM method was selected since it takes advantage of the density functional theory for the accurate treatment of the interactions of reactive intermediates with the Ti site in the crystal framework of TS-1 and of the universal force fields (UFF) for manifestation of the van der Waals interaction due to the confinement of the extended zeolite structure.41-43 This method has been adopted to evaluate the effects of the zeolite framework and also to evaluate the energy profile throughout the reaction mechanisms in this work.

The defect TS-1 was represented by the 65T cluster model, TiSi64O97H74, selected from the ZSM-5 zeolite lattice.44 The number of T atoms refers to tetrahedrally coordinated Ti and Si atoms in the framework model. This model is considered to be sufficiently large to cover all important framework effects acting on both the active site and the adsorbate. It contains a nearly circular straight channel with dimensions of 5.4 × 5.6 Å and a slightly elliptical zigzag channel with dimensions of 5.1 × 5.5 Å. Both channels intersect each other at the middle of the model, thus generating an intersecting channel with the spatial dimension of about 9 Å. The Si atom at site T5 was removed from the lattice framework to model the defect site and the neighboring Si atom at site T6 was replaced with the Ti atom to act as the active center. This model corresponds to constant wavelength powder neutron diffraction data collected on isotopic TS-1 samples which determine both titanium occupying and silicon vacancy sites.21-22 The dangling bonds resulting from the elimination of the Si atom were terminated with hydrogen atoms. This cluster model is illustrated in Figure 2b. The small active region is usually the only region treated accurately with the ab initio method, to conserve computational resources and limit time consumption. However, if more accurate results are required, the effect from the framework structure of zeolite has to be taken into consideration. To address this situation, the two-layered ONIOM scheme was adopted to model the molecular properties of TS-1 and interactions of its complexes with H2O2 and small olefins. The two-layered ONIOM approach enables simplification of the calculation of energies by treating the active region with a high-level quantum mechanical (ab initio or density functional) approach and the extended framework environment with a less expensive level, the HF and molecular mechanics force fields. The total energy of the whole system can be expressed within the framework of the ONIOM methodology developed by Morokuma and his coworkers45

EONIOM ) ELowReal + (EHighCluster - ELowCluster)

(1)

where the superscript “Real” means the whole system and the superscript “Cluster” means the active region. Subscripts “High” and “Low” mean high- and low-level methodologies used in the ONIOM calculation. In this study, the 9T/65T two-layered ONIOM scheme was employed. An unconstrained 9T cluster, which is referred to as “high level region,” was calculated with density functional theory with the hybrid functional B3LYP, while the rest, which is referred to as “low level region,” was treated by the UFF force field to reduce computational time and to practically represent the confinement effect of the zeolite pore structure. This combination scheme has been tested on the 20T cluster model to study the adsorption of ethylene, benzene, and ethylbenzene on the acidic H-FAU zeolite and it has been found that the adsorption energies compared favorably with the full HF and B3LYP levels of theory.41 Gaussian 03 software was used to perform all calculations in this study. The Los Alamos LANL2DZ46 effective core pseudopotentials (ECP) and valence double-ζ basis set were used for the titanium atom and the full double-ζ basis set, 6-31G(d,p), was applied for carbon, hydrogen, oxygen and silicon atoms. During the geometry optimization, only the quantum region, 9T cluster, was allowed to relax while the remainders are fixed at the lattice positions. The transition states, which have only one imaginary frequency whose mode corresponds

Epoxidation Reaction of Unsaturated Hydrocarbons

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3435 TABLE 1: Selected Optimized Structural Parameters of the TS-1 Active Site Model Calculated at the Two-layered ONIOM(B3LYP/6-31G(d,p):UFF) Level of Theory parameter bond distance (Å) Ti-O1 Ti-O2 Ti-O3 Ti-O4 angle (°) Ti-O2-Si2 Ti-O3-Si3 Ti-O4-Si4 a

Figure 2. (a) Two tetrahedral sites (T5 and T6) of active site without defect and (b) optimized structure of defect active site model using the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) method. The highlevel region is displayed with balls and sticks and the low level region is denoted by lines.

to the designated reaction coordinate, were verified by the normal-mode analyses. The minimized geometries were used as inputs in the single point energy calculations at the same level of theory using the 30T/65T two-layered ONIOM scheme to improve the energetic properties. The optimized structural parameters of the TS-1 active site model are given in Table 1. The Ti-O bond lengths are in the range of 1.753-1.820 Å, which is close to the experimental values, 1.793 ( 0.007 Å based on XRD32 and 1.81 ( 0.01 Å obtained by EXAFS,47-49 whereas the Ti-O1(H) bond length is 1.912 Å. The optimized TS-1 structure model using the twolayered ONIOM scheme is shown in Figure 2b, where the high level region is displayed with balls and sticks and the low level region is denoted by lines. 3. Results and Discussion 3.1. Adsorption of H2O2 on the Ti-Substituted Active Site. With the introduction of a H2O2 molecule to the defect active

ONIOM

expt.

1.912 1.820 1.760 1.753 1.778

1.793 ( 0.007,a 1.81 ( 0.01b

157.5 168.8 153.1

From ref 32. b From refs 47-49.

site of TS-1, it interacts with the substituted Ti active center via one (OR) of the two O atoms (OR and Oβ) as shown in Figure 3a. The interaction between OR and the Ti center resembles a coordinative bond with a Ti‚‚‚OR distance of 2.378 Å. Additionally, the H2O2 molecule can form two H-bonds with the local active site O atoms through its H atoms (HR and Hβ). One is a strong H-bond between HR and a silanol Oz atom (Oz‚‚‚ HR ) 1.618 Å), while the other is only a weak interaction between Hβ and an O4 atom of the TS-1 framework (O4‚‚‚Hβ ) 1.904 Å). The insertion of H2O2 causes the geometrical parameters of both the active site and the adsorbing molecule to distort from their unperturbed geometries. The Ti-O(Si) bond, averaged over the three O framework atoms, is elongated by 0.022 Å, whereas the titanol Ti-O1 bond is lengthened by 0.013 Å. The O-O bond of the H2O2 molecule is shortened by 0.015 Å (from 1.456 Å to 1.441 Å), while the O-H bond distance is split from 0.970 Å, an equivalent value of the bare H2O2, to 1.008 Å and 0.978 for the OR-HR and Oβ-Hβ bond distances, respectively. The longer O-H bond distance of H2O2 in the adsorption complex refers to the stronger H-bond formation. With attention to the geometry of the Ti active center, the coordination of an active Ti can be seen to change from a distorted tetrahedral structure in the bare TS-1 to a distorted octahedral conformation in the adsorption complex (Figure 3b). The octahedral conformation involves three Ti-O-Si bridging oxygens, titanol oxygen, a neighboring silanol oxygen, and one of the two oxygens of H2O2. This configuration does not exist when using a nondefect model owing to the restriction of the zeolite framework. The computed adsorption energy is exothermic by 11.6 kcal/ mol, which is much lower than that obtained in the previous theoretical small cluster study at the BPW91 functional of 17 kcal/mol.35 The latter high interaction energy is attributed to the use of a small quantum cluster that does not take into consideration the zeolite framework constraints. Our present study determined that the small 9T quantum cluster calculation leads to the highly exothermic adsorption energy of 19.4 kcal/ mol. When comparing our two-layered ONIOM result with previous theoretical studies, it is found that the defective model gives higher exothermic adsorption energy (11.6 kcal/mol) than that of the fully tetrahedral Ti center models, where the adsorption energy was exothermic by about 7.9-9.1 kcal/ mol.19,27 This is a result of the relatively high flexibility of the Ti active site that allows it to accommodate lower energy configurations of the adsorption complex. In addition, the presence of the neighboring silanol groups resulting from a Si vacancy provides a substantial stabilization of the adsorption complex through H-bond formation.

3436 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Panyaburapa et al.

Figure 3. (a) The optimized structure of adsorption complex (Ads_1) of TS-1 and H2O2 calculated at the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) method and (b) its close-up distorted octahedral conformation. For clarity, some of the quantum region (balls and sticks) and the rest of the UFF region (lines) are omitted.

3.2. Oxidative Active Site Formation. Although, theoretically, the physisorption of H2O2 on the Ti active center is observable11,13,35 and has been reported in literature, there has not, so far, been any experimental report published of an undissociated adsorption form. Instead, the dissociated adsorption known as the titanium hydroperoxo species, Ti-OOH, is undoubtedly detected after TS-1 has been in contact with the H2O2/H2O solution.19,50 It is understood that this species is the active oxidizing intermediate in the Ti/H2O2 system. However, no intensive study has been made of an elementary step of the oxidative active site formation. The Ti-OOH formation through the direct proton transfer from H2O2 to the titanol OH group with the calculated activation energy of 15.4 kcal/mol has recently been proposed by Wells and co-workers,35 while Sever and Root have shown that the Gibbs activation energy for TiOOH formation is lowered by 5.0-6.0 kcal/mol when a protic solvent molecule is present in the hydrogen-bonded bridging position.13 In our present study, which considers the physisorption complex (Figure 3), it is evident that the Ti-OOH formation can be accomplished in two ways, either via a direct single proton transfer or an indirect double proton-transfer mechanism. The Ti-OOH formation can occur when a proton of H2O2 is directly attached to the titanol OH group in the onestep single proton-transfer mechanism, or, in the one-step double proton-transfer mechanism, when it is transferred to the proximal silanol OH group, which serves as a proton donor-acceptor group. Figure 4a shows the transition state structure of a single proton-transfer mechanism. The imaginary vibration mode obtained from the frequency calculation corresponds to the protonation of the titanol group. The HR of H2O2 migrates directly to the titanol OH group while, simultaneously, the OR is coordinating with the Ti active center. The OR-HR bond breaks and is lengthened to 1.186 Å, whereas the Ti‚‚‚OR separation is contracted to 2.336 Å. In this process, a poor leaving group (OH-) is converted into a good one (H2O). Hence, a water molecule is being developed during the formation of an oxidative active species. The activation energy for the direct proton transfer is estimated to be 10.8 kcal/mol. Figure 4b shows the protonation taking place through a double proton-transfer mechanism with the assistance of a nearby silanol group, a transition state structure that appears more

Figure 4. Optimized geometrical structures of (a) single proton transfer and (b) double proton-transfer mechanisms calculated at the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) level of theory.

complicated than the single proton transfer. Instead of passing HR directly to the titanol OH group, the H2O2 transfers its HR to the neighboring silanol OH group, which behaves like a bridge linking H2O2 and a titanol group. The normal coordinate associated with the imaginary frequency shows the synchronized dual proton-transfer mechanism where two protons, one from H2O2 and another from silanol OH group, move simultaneously toward the partial negative-charge oxygen atoms (Oz and O1). This mechanism reduces the activation energy to 2.6 kcal/mol. In the double proton-transfer process, the pre-hydrogen-bonded system increases the degree of proton transfer, whereas, in the single proton-transfer mechanism, it is necessary for one of the two hydrogen bonds in the adsorption complex to break before the protonation. Therefore, in the defect active site model of TS-1, the formation of an oxidative active species preferentially occurs through the double proton-transfer mechanism. 3.3. Stability of the Oxidative Active Species. The completion of the proton-transfer process takes place with the formation of a water molecule and the titanium hydroperoxo complexes (Ti-OOH). These species, which can be formed either in a

Epoxidation Reaction of Unsaturated Hydrocarbons

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3437

Figure 5. Optimized geometrical structures of Ti-hydroperoxo complexes in different conformations calculated at the two-layered ONIOM(B3LYP/ 6-31G(d,p):UFF) level of theory.

mono (η1) or bidentate (η2) form, have been modeled as the oxidative active species.25,35 The environmental surroundings as shown in Figure 5 determine their stability with respect to the adsorption complex (Ads_1). The monodentate structure (Int_1), the first Ti-OOH species formed after H2O2 contacted with the Ti active center, is the most stable configuration being more stable than an undissociated adsorption complex, Ads_1, by 10.3 kcal/mol. Its hydrated form allows the Ti active center to form a possibly stable octahedral structure with the coordination number of six (Figure 6), larger than that of the bidentate structures (Int_2, Int_3, and Int_4) of five. It is evident that the stability of Ti-OOH species is determined not only by the coordination number of the Ti atom, but also by the number of strong hydrogen bonds. For the bidentate structures, the Int_4 is the most stable in comparison to Int_2 and Int_3 configurations. It is more stable than the physisorbed complex of H2O2 (Ads_1) by 5.3 kcal/mol. This results from its Hβ being able to form a strong hydrogen bond with the water molecule. For the Int_2 and Int_3 structures where the acidic hydrogen (Hβ) does not directly form an H-bond with the water molecule, their

stability is relatively low as compared to the Int_4 complex. With respect to the Ads_1 complex, the Int_3 configuration is slightly more stable (by 2.3 kcal/mol), whereas the Int_2 was found to be the least stable structure. It is less stable than the Ads_1 complex by 5.4 kcal/mol. Titanium peroxo species, Ti-(OO-), in the η2-configuration is the only one that can be detected experimentally and its existence has been confirmed by the Raman spectroscopic study of the yellow color of the TS-1/H2O/H2O2 system, which shows the frequency at 618 cm-1.19-20 Despite there being no evidence that the Ti-(OO-) is an active species in the partial oxidation reactions, it can be in equilibrium with the hydroperoxo complexes in the presence of water. It is possible that the peroxo complex is formed by the evolution of both η1 and η2 hydroperoxo species in the presence of water molecules via the formation of H3O+/H2O. Figure 7 shows the Ti peroxo complex, Int_5, in bidentate manner. This species is found to be more stable than the Ads_1 by 3.4 kcal/mol. Two O atoms of H2O2 coordinate to Ti with distances of 1.934 and 1.944 Å, and the O-O bond is lengthened from 1.445 Å of an isolated H2O2 to

3438 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Figure 6. Close-up of the distorted octahedral structure of Int_1 or Ti-(η1-OOH).

Figure 7. Optimized geometrical structure of the Ti-peroxo complex calculated at the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) level of theory.

1.456 Å. One of the O atoms forms a hydrogen bond with the next silanol group while the other is stabilized by H2O and H3O+, which form a hydrogen bond to each other. It is noted that, by using the 9T/65T two-layered ONIOM scheme, yet in the absence of an additional water molecule, the peroxo complex could not be formed. This finding is in accordance with the role of water in hydroperoxo-peroxo interconversion in the TS1/H2O/H2O2 system.19 3.4. Epoxidation of Unsaturated Hydrocarbons. A range of stable intermediates have been proposed to be active species in the epoxidation of unsaturated hydrocarbons.17,24-27,51-53 These include titanium hydroperoxo complexes, Ti-(OOH), in mono- and bidentate structures as well as the titanium peroxo complex, Ti-(OO-). Even though the peroxo complex plays a role in the Ti-peroxo and Ti-hydroperoxo interconversion in the aqueous solution, it exhibits poor catalytic activity toward partial oxidation reactions, which is why we focus only on the Tihydroperoxo species as oxidative active sites in this study. Only the bidentate structures are considered to be involved in the partial oxidation because the monodentate form is too compact for the olefins to access. Our calculations show that Int_3 and Int_4 are the energetically most stable configurations of the bidentate structures. These two different conformations enable us to investigate the ethylene attack at OR of hydroperoxo species in the cross section and straight channel of the zeolite pore for Int_3, and Int_4, respectively. The lack of knowledge pertaining to the ethylene attack at the Oβ mechanism has been

Panyaburapa et al. expounded by various previous studies. Tantanak and coworkers have shown that the activation barrier for the ethylene attack at the proximal OR atom is about 20.0 kcal/mol, lower than that at the distal Oβ atom.26 Yudanov investigated a distal attack mechanism which involved the concerted cleavage of the Oβ-OR bond and proton transfer from Oβ to OR. The activation barrier was found to exceed the barrier for the proximal attack by about 13.0 kcal/mol.54 Using the unconstrained first shell coordination titanium hydroperoxo cluster, including solvent effects, Sever demonstrated that the ethylene attack at the proximal OR atom reduced the Gibbs free energy changes by 10.0 kcal/mol.12 Both Limtrakul et al.25 and Thomson et al.35 have similarly proposed the proximal attack for the epoxidation of ethylene and propylene, but without solvent. We have, therefore, considered only the ethylene attack at OR in the oxygen abstraction step. When the titanium hydroperoxo complex has been formed, it can rapidly interact with ethylene either via the OR or Hβ site. Even though, we were unable to locate the stable ethylene adsorbed on OR of Ti(η2-OOH) due to the weak repulsion between OR and the π bond of ethylene. The ethylene molecule preferred to reside away from the Ti-hydroperoxo oxygenated species, which agrees well with previous theoretical studies using small cluster models, where ethylene prefers to interact with Ti(η2-OOH) by forming the hydrogen bond with the OβHβ moiety with adsorption energies of 0.9-2.6 kcal/mol.26 We have not proposed this adsorption complex in our pathway since the conformations of the adsorption complex do not allow for the oxygen abstraction. Moreover, this process can be overcome with the addition of some energy to break the hydrogen bond between ethylene and the Oβ-Hβ group. We stated earlier that there are two regions where the epoxidation takes place: at the cross section and in the straight channel of the zeolite pore. Figure 8a,b shows two possible locations of transition state complexes of the ethylene epoxidation. It is found that the epoxidation of ethylene in the straight channel requires a much higher energy (6.0 kcal/mol) to overcome the activation barrier as compared to that at the cross section. This is attributed to the zeolite framework constraint of the straight channel which becomes congested when ethylene is introduced. The activation barrier of epoxidation in the straight channel is certainly expected to be larger when bulky olefin molecules are present. Consequently, in this study, we anticipate that only the cross section regions are able to accommodate all bulky transition state complexes in the epoxidation pathway. The apparent activation energy for the ethylene epoxidation is calculated to be 15.5 kcal/mol, which is in good agreement with the work of Limtrakul and co-workers for the ethylene epoxidation (15.3 kcal/mol) on the small quantum cluster embedded in the Madelung potential of the zeolite framework.25 Sever and Root reported the calculated activation barrier of the ethylene epoxidation of 14.5 kcal/mol by using the first shell unconstrained Ti(η2-OOH) cluster with one water ligand.12 Even though we are unable to procure experimental data on the ethylene epoxidation for direct comparison, it is encouraging that our apparent activation energy is comparable to the related work of Langhendries and co-workers, who published the apparent activation energy of 15.5 ( 1.5 kcal/mol for the epoxidation of 1-hexene in the TS-1/CH3OH/H2O2 system.55 The transition state complex of the ethylene epoxidation (ts_EO1) involves the proximal OR abstraction of the ethylene π bond and the OR-Oβ bond dissociation (Figure 8). With respect to the Int_4, the Ti-OR bond is lengthened from 1.934 to 1.998 Å, whereas the Ti-Oβ bond is shortened from 2.303

Epoxidation Reaction of Unsaturated Hydrocarbons

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3439

Figure 8. Optimized transition state structures of the ethylene epoxidation located at (a) the cross section and (b) the straight channel. All optimized parameters are obtained from the energy minimization at the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) level of theory. Some parts of the quantum region (balls and sticks) and of MM region (lines) are omitted for clarity.

to 2.041 Å. These results correspond to the OR-Oβ bond elongation from 1.455 to 1.795 Å. An approach of ethylene to the OR results in the CdC bond lengthening from 1.330 to 1.359 Å. The C-O distance is split into 1.992 and 2.260 Å, suggesting that the epoxide formation is an asynchronous type mechanism. The overall oxidation pathway of ethylene is shown in Figure 9. From the energy profile of the ethylene epoxidation, it can be seen that the O abstraction step is the overall rate determining step (Ea ) 32.5 kcal/mol). To determine the effect of the hydrocarbon chain length on the oxidative reactivity of TS-1, by using propylene and trans2-butylene, additional investigation was undertaken. The transition state structures of propylene and trans-2-butylene shown in Figure 10 are similar to that of ethylene, and both involve the olefin attack at the proximal OR of the Ti(η2-OOH) complex. Despite the geometrical structures appearing comparable to the ethylene epoxidation, the apparent activation barriers are different. The apparent activation energy for propylene is estimated to be 13.6 and 12.2 kcal/mol for trans-2-butylene, (Figure 9), less than that of ethylene. Attention is drawn to the fact that the activation barrier decreases with increasing the chain length and the number of methyl groups. This trend is consistent with the gas-phase calculations of the ethylene, propylene and trans2-butylene epoxidation with H2O2 at the same level of theory. As the substitution of the methyl group, which is a weak electron-donating group, to the active CdC bond leads to the

increase of the nucleophilicity of olefinic hydrocarbons, the reactivity series for the O abstraction is, therefore, in the order ethylene < propylene < trans-2-butylene with the apparent activation energies of 36.3, 34.3, and 32.8 kcal/mol, respectively. The reaction energies of the epoxide formation (Figure 9) are exothermic by 44.8, 46.3, and 63.6 kcal/mol for ethylene, propylene, and trans-2-butylene, respectively. 4. Conclusions The partial oxidation of unsaturated hydrocarbons with H2O2 over the defect active site of TS-1 catalyst has been systematically studied by using the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) method. The reaction starts with the adsorption of a hydrogen peroxide molecule on the Ti atom (Ads_1) via the H-bond formation with a neighboring silanol group and a coordinative interaction with the Ti atom. This is followed by the oxidation pathway of alkenes proceeded through (i) the formation of an oxidative active site on the Ti atom of TS-1 and (ii) the alkene oxidation, where the former is associated with a proton transfer from the hydrogen peroxide to the OH group of the Ti active center to simultaneously develop a water molecule and titanium hydroperoxo intermediates. Two possible mechanisms of this step were investigated: a direct proton transfer and a double proton transfer aided by the neighboring silanol group. The investigation showed that the absence of a

3440 J. Phys. Chem. C, Vol. 111, No. 8, 2007

Panyaburapa et al.

Figure 9. Schematic energy profile of the overall epoxidation reaction of ethylene, propylene, and trans-2-butylene with H 2O2 over TS-1 calculated at the ONIOM(B3LYP/6-31G(d,p):UFF) using the 30T/65T two-layered ONIOM scheme.

Figure 10. Optimized transition state complexes of (a) propylene and (b) trans-2-butylene epoxidation with H2O2 over TS-1 calculated at the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) level of theory.

Si atom adjacent to the Ti active center promisingly improves the agreement of existing evidence for the formation of oxidative active sites and their stability. The double proton-transfer mechanism (ts_2) was preferred over the direct proton transfer (ts_1). The activation energies for the double and direct proton transfer were calculated to be 2.6 and 10.8 kcal/mol, respectively. The titanium hydroperoxo intermediate (Int_4) in the bidentate form, Ti(η2-OOH), was selected as the active species in the alkene oxidation process and served as an oxygen atom donor. This species was converted to the titanium peroxo (Int_5), Ti(η2-OO-) complex when an additional molecule was added. In the latter step, alkene molecules were introduced to the active site to be oxidized to the epoxide product. The results showed that the oxygen abstraction step was the rate determining step of the alkene oxidation reaction, and that the energy barrier of this step decreased with increasing the hydrocarbon chain length and the number of methyl groups. Therefore, the nucleophilicity of the alkene molecule was the considerable effect that it had on the activation energy of the oxidation reaction. The apparent activation energies for ethylene, propy-

lene, and trans-2-butylene were predicted to be 15.5, 13.6, and 12.2 kcal/mol, respectively, which agreed well with the experimental epoxidation of 1-hexene in the TS-1/CH3OH/H2O2 system of 15.5 ( 1.5 kcal/mol. These results substantiate that among the sophisticated hybrid techniques for investigating many reactions in zeolite catalysts, the two-layered ONIOM(B3LYP/6-31G(d,p):UFF) scheme is a good method of choice. Acknowledgment. This work was supported in part by grants from the Thailand Research Fund (TRF Senior Research Scholar to JL.), the National Nanotechnology Center (NANOTEC, Thailand), the Kasetsart University Research and Development Institute (KURDI), and the Ministry of University Affairs under the Science and Technology Higher Education Development Project (MUA-ADB funds). References and Notes (1) Van Santen, R. A.; Kramer, G. J. Chem. ReV. 1995, 95, 637. (2) Notari, B. AdV. Catal. 1996, 41, 253. (3) Clerici, M. G.; Bellussi, G.; Romano, U. J. Catal. 1991, 129, 159.

Epoxidation Reaction of Unsaturated Hydrocarbons (4) Mantegazza, M. A.; Petrini, G.; Fornasari, G.; Rinaldo, A.; Trifiro, F. Catal. Today 1996, 32, 297. (5) Khouw, C. B.; Dartt, C. B.; Labinger, J. A.; Davis, M. E. J. Catal. 1994, 149, 195. (6) van der Pol, A. J. H. P.; van Hooff, J. H. C. Appl. Catal. A 1993, 106, 97. (7) Kumar, R.; Bhaumik, A. Micropor. Mesopor. Mater. 1998, 21, 497. (8) Bengoa, J. F.; Gallegos, N. G.; Marchetti, S. G.; Alvarez, A. M.; Cagnoli, M. V.; Yeramian, A. A. Micropor. Mesopor. Mater. 1998, 24, 163. (9) Zecchina, A.; Bordiga, S.; Lamberti, C.; Ricchiardi, G.; Scarano, D.; Petrini, G.; Leofanti, G.; et al. Catal. Today 1996, 32, 97. (10) Damin, A.; Bordiga, S.; Zecchina, A.; Lamberti, C. J. Chem. Phys. 2002, 117, 226. (11) Sinclair, P. E.; Catlow, C. R. A. J. Phys. Chem. B 1999, 103, 1084. (12) Sever, R. R.; Root, T. W. J. Phys. Chem. B 2003, 107, 4090. (13) Sever, R. R.; Root, T. W. J. Phys. Chem. B 2003, 107, 4080. (14) Tozzola, G.; Mantegazza, M. A.; Ranghino, G.; Petrini, G.; Bordiga, S.; Ricchiardi, G.; Lamberti, C.; Zulian, R.; Zecchina, A. J. Catal. 1998, 179, 64. (15) Barker, C. M.; Gleeson, D.; Kaltsoyannis, N.; Catlow, C. R. A.; Sankar, G.; Thomas, J. M. Phys. Chem. Chem. Phys. 2002, 4, 1228. (16) Sankar, G.; Thomas, J. M.; Catlow, C. R. A.; Barker, C. M.; Gleeson, D.; Kaltsoyannis, N. J. Phys. Chem. B 2001, 105, 9028. (17) Sinclair, P. E.; Sankar, G.; Catlow, C. R. A.; Thomas, J. M.; Maschmeyer, T. J. Phys. Chem. B 1997, 101, 4232. (18) Zhuang, J.; Ma, D.; Yan, Z.; Deng, F.; Liu, X.; Han, X.; Bao, X.; Liu, X. W.; Guo, X.; Wang, X. J. Catal. 2004, 221, 670. (19) Bonino, F.; Damin, A.; Ricchiardi, G.; Ricci, M.; Spano, G.; D’Aloisio, R.; Zecchina, A.; Lamberti, C.; Prestipino, C.; Bordiga, S. J. Phys. Chem. B 2004, 108, 3573. (20) Bordiga, S.; Damin, A.; Bonino, F.; Ricchiardi, G.; Zecchina, A.; Tagliapietra, R.; Lamberti, C. Phys. Chem. Chem. Phys. 2003, 5, 4390. (21) Lamberti, C.; Bordiga, S.; Zecchina, A.; Artioli, G.; Marra, G.; Spano, G. J. Am. Chem. Soc. 2001, 123, 2204. (22) Henry, P. F.; Weller, M. T.; Wilson, C. C. J. Phys. Chem. B 2001, 105, 7452. (23) Hijar, C. A.; Jacubinas, R. M.; Eckert, J.; Henson, N. J.; Hay, P. J.; Ott, K. C. J. Phys. Chem. B 2000, 104, 12157. (24) Atoguchi, T.; Yao, S. J. Mol. Catal. A 2003, 191, 281. (25) Limtrakul, J.; Inntam, C.; Truong, T. N. J. Mol. Catal. A 2004, 207, 139. (26) Tantanak, D.; Vincent, M. A.; Hillier, I. H. Chem. Commun. 1998, 1031. (27) Vayssilov, G. N.; Van Santen, R. A. J. Catal. 1998, 175, 170. (28) Boccuti, M. R.; Rao, K. M.; Zecchina, A.; Leofanti, G.; Petrini, G. Stud. Surf. Sci. Catal. 1989, 48, 133. (29) Munakata, H.; Oumi, Y.; Miyamoto, A. J. Phys. Chem. B 2001, 105, 3493. (30) Bellussi, G.; Carati, A.; Clerici, M. G.; Maddinelli, G.; Millini, R. J. Catal. 1992, 133, 220.

J. Phys. Chem. C, Vol. 111, No. 8, 2007 3441 (31) Bordiga, S.; Coluccia, S.; Lamberti, C.; Marchese, L.; Zecchina, A.; Boscherini, F.; Buffa, F.; Genoni, F.; Leofanti, G.; Petrini, G.; Vlaic, G. J. Phys. Chem. 1994, 98, 4125. (32) Lamberti, C.; Bordiga, S.; Arduino, D.; Zecchina, A.; Geobaldo, F.; Spano, G.; Genoni, F.; Petrini, G.; Carati, A.; Villain, F.; Vlaic, G. J. Phys. Chem. B 1998, 102, 6382. (33) Ratnasamy, P.; Srinivas, D.; Kno¨zinger, H. AdV. Catal. 2004, 48, 1. (34) To, J.; Sherwood, P.; Sokol, A. A.; Bush, I. J.; Catlow, C. R. A.; van Dam, H. J. J.; French, S. A.; Guest, M. F. J. Mater. Chem. 2006, 16, 1919. (35) Wells, D. H., Jr.; Delgass, W. N.; Thomson, K. T. J. Am. Chem. Soc. 2004, 126, 2956. (36) Shah, R.; Gale, J. D.; Payne, M. C. J. Phys. Chem. B 1997, 101, 4787. (37) Jeanvoine, Y.; Angyan, J. G.; Kresse, G.; Hafner, J. J. Phys. Chem. B 1998, 102, 5573. (38) Hillier, I. H. J. Mol. Struct. (THEOCHEM) 1999, 463, 45. (39) Dungsrikaew, V.; Limtrakul, J.; Hermansson, K.; Probst, M. Int. J. Quantum Chem. 2003, 96, 17. (40) Sauer, J.; Sierka, M.; J. Comput. Chem. 2000, 21, 1470. (41) Kasuriya, S.; Namuangruk, S.; Treesukol, P.; Tirtowidjojo, M.; Limtrakul, J. J. Catal. 2003, 219, 320. (42) Namuangruk, S.; Pantu, P.; Limtrakul, J. J. Catal. 2004, 225, 523. (43) Jansang, B.; Nanok, T.; Limtrakul, J. J. Phys. Chem. B 2006, 110, 12626. (44) van Koningsveld, H.; van Bekkum, H.; Jansen, J. C. Acta. Crystallogr. 1987, B43, 127. (45) Svensson, M.; Humbel, S.; Froese, R. D. J.; Matsubara, T.; Sieber, S.; Morokuma, K. J. Phys. Chem. 1996, 100, 19357. (46) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 299. (47) Pei, S.; Zajac, G. W.; Kaduk, J. A.; Faber, J.; Boyanov, B. I.; Duck, D.; Fazzini, D.; Morrison, T. I.; Yang, D. S. Catal. Lett. 1993, 21, 333. (48) Davis, R. J.; Liu, Z.; Tabora, J. E.; Wieland, W. S. Catal. Lett. 1995, 34, 101. (49) Bordiga, S.; Coluccia, S.; Lamberti, C.; Marchese, L.; Zecchina, A.; Boscherini, F.; Buffa, F.; Genoni, F.; Leofanti, G.; et al. J. Phys. Chem. 1994, 98, 4125. (50) Lin, W.; Frei, H. J. Am. Chem. Soc. 2002, 124, 9292. (51) Mimmoun, H.; Postel, M.; Casabianca, F.; Fisher, J.; Mitschler, A. Inorg. Chem. 1982, 21, 1303. (52) Yamase, T.; Ishikawa, E.; Asai, Y.; Kanai, S. J. Mol. Catal. A 1996, 114, 237. (53) Feixue, G.; Toshihiro, Y.; Hideo, S. J. Mol. Catal. A 2002, 180, 97. (54) Yudanov, I. V.; Gisdakis, P.; Di Valentin, C.; Rosch, N. Eur. J. Inorg. Chem. 1999, 2135. (55) Langhendries, G.; De Vos, D. E.; Baron, G. V.; Jacobs, P. A. J. Catal. 1999, 187, 453.